1040049 - [dolphin][Window Management] when close an app, the app zoom-out animation doesn't seems smoothly

Status: RESOLVED WONTFIX

(Firefox OS Graveyard :: Gaia::System::Window Mgmt, defect, P1)

Description

[yong.ren]

DEFECT DESCRIPTION:
  when close an app, the app zoom-out animation doesn't seems smoothly

Steps to reproduce:
----------------------------------------------------
1. open an app from homescreen, for example email app. 
2. wait for the app stated successfully
3. press the home button 

Actual result:
---------------------------------------
  the app zoom-out animation doesn't seems smoothly.
 

Expected result:
--------------------------------------- 
  the app zoom-out animation should seems smoothly 


Additional info:
--------------------------------------
Reproduce rate:  5/5

Comment 1

[yong.ren]

compare to Firefox OS v1.3, I found the 1.4 zoom-out animation time-function data type style have changed from ease to cubic-bezier function. 

and 1.3 zoom-out animation seemed smoothly, and I don't know why changed the time-function in firefox os 1.4.

Comment 2

[Evelyn Hung]

ni alive to answer why we changed the animation time-function.

Comment 3

[Evelyn Hung]

So Tim suggests we run a layout and graphic profile to find out the potential issue. Time-function change seems not the root cause here. 

Clear all ni flag, and ni Peter to help on profiling. Thanks.

Comment 4

[Ivan Tsay (:ITsay)]

this is blocker according to its symptom.

Comment 5

[Peter Chang[:pchang]]

ni solomon for profiling.
But alive, we would like to know the difference of animation part between v1.3 and v1.4 first.

Comment 6

[Tim Guan-tin Chien [:timdream] (please needinfo)]

Peter, I am fairly certain the timing function change was intended to hide the initial delay, but again I don't think that's the root cause of this bug.

Comment 7

[Peter Chang[:pchang]]

(In reply to Tim Guan-tin Chien [:timdream] (MoCo-TPE) (please ni?) from comment #6)
> Peter, I am fairly certain the timing function change was intended to hide
> the initial delay, but again I don't think that's the root cause of this bug.

Tim, I assume the following are the places to describe the close animation for app. Am I right?
v 1.3
https://github.com/mozilla-b2g/gaia/blob/v1.3/apps/system/style/window.css#L103
v 1.4
https://github.com/mozilla-b2g/gaia/blob/v1.4/apps/system/style/window.css#L211

Solomon, please help to check the gecko profiling data and try to modify gaia as the same behavior to see any difference.

Comment 8

[Tim Guan-tin Chien [:timdream] (please needinfo)]

I think so, but Alive can double confirm.

Comment 10

[(inactive after 6/18) Alive Kuo [:alive]]

No. The animation for app close is defined:
For v1.3
https://github.com/mozilla-b2g/gaia/blob/v1.3/apps/system/style/system/system.css#L482
For v1.4
https://github.com/mozilla-b2g/gaia/blob/v1.4/apps/system/style/window.css#L285

The timing function is
For v1.3
https://github.com/mozilla-b2g/gaia/blob/v1.3/apps/system/style/system/system.css#L408
For v1.4
https://github.com/mozilla-b2g/gaia/blob/v1.4/apps/system/style/window.css#L211

Comment 11

[Solomon Chiu [:schiu]]

We just compared the difference animation transition function, with 2 devices which applied the transition functions(ease v.s. cubic-bezier) respectively. By visually checking the devices side by side, they seems don't have significant different on smoothness. 

Yon.ren:

Could you please record and provide the video that with the comparision of smothness between 1.3 and 1.4 devices, such that we can inspect the smothness difference that you observed?

Thanks.

Comment 12

[yong.ren]

Solomon:

   I have already take the 1.3 and 1.4 device's video, but it's too big(18M each), can't attach on this bug. Could you please give me a path to submit the video, thanks.

Comment 13

[yong.ren]

Just as comment 12

Comment 14

[yong.ren]

due to comment 12's reason, I (In reply to Solomon Chiu [:schiu] from comment #11)
> We just compared the difference animation transition function, with 2
> devices which applied the transition functions(ease v.s. cubic-bezier)
> respectively. By visually checking the devices side by side, they seems
> don't have significant different on smoothness. 
> 
> Yon.ren:
> 
> Could you please record and provide the video that with the comparision of
> smothness between 1.3 and 1.4 devices, such that we can inspect the
> smothness difference that you observed?
> 
> Thanks.

Due to Comment 12's reason, I have send the video to your email, please confirm when you receive it. thanks.

Comment 15

[Solomon Chiu [:schiu]]

Hi Yong.ren,

I am checking with my colleagues to see if there is available ftp site. Meanwhile could you please use dropbox or other kind of cloud storage to upload the video first?

Thanks.

Comment 16

[yong.ren]

(In reply to Solomon Chiu [:schiu] from comment #15)
> Hi Yong.ren,
> 
> I am checking with my colleagues to see if there is available ftp site.
> Meanwhile could you please use dropbox or other kind of cloud storage to
> upload the video first?
> 
> Thanks.

Please see the comment 14, I([e-mail]) have send the video to your email.  please confirm when you receive it, thanks.

Comment 17

[Solomon Chiu [:schiu]]

Just checked the video files you provided, and found the comparison is made with Tarako and Dolphin. There is a considerable factor which impacting smoothness is the panel size - Tarako uses a 320x480 panel, and Dolphin uses a 480x854 panel. The panel size of Dolphin is 2.67 times of Tarako's. I think the comparing should be made on the same devices(Dolphin with 1.3 and 1.4) for a fair comparison.

Comment 18

[James Zhang (Spreadtrum)]

1.3 don't support android4.4.

Comment 19

[Peter Chang[:pchang]]

[Blocking Requested - why for this release]:

(In reply to Solomon Chiu [:schiu] from comment #17)
> Just checked the video files you provided, and found the comparison is made
> with Tarako and Dolphin. There is a considerable factor which impacting

I think the performance comparison between Tarako and Dolphin is not fair because different resolution size.

Things we could do here is to check the performance of flame 1.4 by enabling single cpu core.
> smoothness is the panel size - Tarako uses a 320x480 panel, and Dolphin uses
> a 480x854 panel. The panel size of Dolphin is 2.67 times of Tarako's. I
> think the comparing should be made on the same devices(Dolphin with 1.3 and
> 1.4) for a fair comparison.

Comment 20

[Luke Chang [:lchang] (inactive)]

Attachment: dolphin v1.3 vs. v1.4

I've tried the performance with gaia v1.3 and v1.4 on the dolphin and recorded a video as the attachment. I think there's no significant difference on smoothness between these two cases.

In the video, left device is v1.3 and right one is v1.4. Follows are their configurations:

v1.3 (left)

Gaia      23f55be856cef53c6604a6fe4aeb09061afbc897
Gecko     https://hg.mozilla.org/releases/mozilla-b2g30_v1_4/rev/1f7f2fb4ea12
BuildID   20140729160209
Version   30.0
ro.build.version.incremental=89
ro.build.date=Tue May 13 06:09:31 CST 2014

v1.4 (right)

Gaia      eb3b185325901d4c04e2d43eb58d90835213bea9
Gecko     https://hg.mozilla.org/releases/mozilla-b2g30_v1_4/rev/1f7f2fb4ea12
BuildID   20140729160209
Version   30.0
ro.build.version.incremental=74
ro.build.date=Sat Apr 26 07:30:05 CST 2014

Comment 21

[Solomon Chiu [:schiu]]

Attachment: systrace of dolphin-1.4 and flame-1.4

Comment 22

[Solomon Chiu [:schiu]]

Just uploaded the systrace result of dolphin and flame, the test conditions are listed as blow:

1. dolphin 1.4: LCD(480x854), single core
2. flame 1.4: LCD(480x854), dual core with CPU-1 diabled

Comment 23

[Peter Chang[:pchang]]

(In reply to peter chang[:pchang][:peter] from comment #19)
> [Blocking Requested - why for this release]:
> 
> (In reply to Solomon Chiu [:schiu] from comment #17)
> > Just checked the video files you provided, and found the comparison is made
> > with Tarako and Dolphin. There is a considerable factor which impacting
> 
> I think the performance comparison between Tarako and Dolphin is not fair
> because different resolution size.
Hi Wayne,
   According to above comment, I would like to confirm the priority of this bug.
> 
> Things we could do here is to check the performance of flame 1.4 by enabling
> single cpu core.
> > smoothness is the panel size - Tarako uses a 320x480 panel, and Dolphin uses
> > a 480x854 panel. The panel size of Dolphin is 2.67 times of Tarako's. I
> > think the comparing should be made on the same devices(Dolphin with 1.3 and
> > 1.4) for a fair comparison.

Comment 24

[James Zhang (Spreadtrum)]

I don't think the performance issue depend on resolution.
You can also try dolphin with hvga parameter. The performance is also bad.

DEVICE=scx15_sp7715ga
LUNCH=scx15_sp7715gaplushvga-userdebug

Comment 25

[James Zhang (Spreadtrum)]

(In reply to peter chang[:pchang][:peter] from comment #19)
> [Blocking Requested - why for this release]:
> 
> (In reply to Solomon Chiu [:schiu] from comment #17)
> > Just checked the video files you provided, and found the comparison is made
> > with Tarako and Dolphin. There is a considerable factor which impacting
> 
> I think the performance comparison between Tarako and Dolphin is not fair
> because different resolution size.
> 
> Things we could do here is to check the performance of flame 1.4 by enabling
> single cpu core.
> > smoothness is the panel size - Tarako uses a 320x480 panel, and Dolphin uses
> > a 480x854 panel. The panel size of Dolphin is 2.67 times of Tarako's. I
> > think the comparing should be made on the same devices(Dolphin with 1.3 and
> > 1.4) for a fair comparison.

The animation performance of 1.4 dolphin hvga is worse than 1.3t tarako.

Comment 26

[Marvin Khoo [:Marvin_Khoo]]

Hi Solomen,
Could you please help compare with Dolphin vs Buri? thanks!

Comment 27

[Solomon Chiu [:schiu]]

Attachment: layerscope-window-reducing

Luke,

We found that the Homescreen is rendered with 3 desktop contents during the animation process. So we think the animation maybe performed better if there is only one desktop be rendered. Let discuss about if we can improve it by using 
well-change property next Monday.

Comment 28

[Solomon Chiu [:schiu]]

Luke,

Please also see my attachment to check the analysis result of Layerscope.

Comment 29

[Solomon Chiu [:schiu]]

(In reply to Marvin Khoo [:Marvin_Khoo] from comment #26)
> Hi Solomen,
> Could you please help compare with Dolphin vs Buri? thanks!

Buri's LCD resolution is 320*720, I think the comparison is not fair.

Comment 30

[yong.ren]

Attachment: dophin 1.4 animation time-function modify

we have modify the dophin 1.4 appwindow animation time-function from cubic-bezier to ease. and It feels the visual effect is better.

Comment 31

[Wayne Chang [:wchang]]

1.4- as examining this on device the performance is acceptable.

Comment 32

[Luke Chang [:lchang] (inactive)]

Clear NI since this bug is set to be 1.4-.

Just in case, our discovery is that "well-change" property seems not to be able to improve the performance. We tried to leave only one page on the homescreen but it doesn't get better.

Comment 33

[Vivien Nicolas (:vingtetun) (:21) - (NOT reading bugmails, needinfo? please)]

I do wonder if bug 1059164 may help here. If so this is a pretty safe patch.

Comment 34

[BMO Automation]

Firefox OS is not being worked on